Duality methods with an automatic choice of parameters. Application to shallow water equations in conservative form (Q5944731)

From MaRDI portal
Revision as of 10:53, 30 July 2024 by Openalex240730090724 (talk | contribs) (Set OpenAlex properties.)
scientific article; zbMATH DE number 1655037
Language Label Description Also known as
English
Duality methods with an automatic choice of parameters. Application to shallow water equations in conservative form
scientific article; zbMATH DE number 1655037

    Statements

    Duality methods with an automatic choice of parameters. Application to shallow water equations in conservative form (English)
    0 references
    0 references
    0 references
    0 references
    26 February 2002
    0 references
    The authors consider the following elliptic variational inequality: \[ \begin{aligned} &\text{Find }u\in V\text{ such that}:\\ &\langle Au, z-u\rangle+ j(z)- j(u)\geq \langle L,z-u\rangle\qquad\forall z\in V,\end{aligned} \] where \(V\) is a real Banach space, \(A\in L(V,V)\), \(L\in V'\), \(j\) is continuous and convex on \(V\) given in an integral form. In order to apply a numerical algorithm to this problem, they reformulate this problem to a new form containing two parameters \(\lambda\) (coming from the Yoshida approximation) and \(w\). \textit{A. Bermúdez} and \textit{C. Moreno} [Comput. Math. Appl. 7, 43-58 (1981; Zbl 0456.65036)] presented an algorithm, which is convergent under appropriate choice of these parameters. Nevertheless, numerical algorithms show that making a good choice is essential in order to have an acceptable rate of convergence. The authors propose a modification of the Bermúdez-Moreno algorithm, which consists in treating \(\lambda\) and \(w\) as functions of space variable \(x\). First, under an appropriate set of assumptions, the convergence of the method is proved. Next, the discussion on a good choice of the parameters is given, followed by two algorithms computing their optimal values, which in fact depend on the solution. Finally, some numerical examples are given, showing good performance of the proposed algorithms.
    0 references
    duality method
    0 references
    maximal monotone operator
    0 references
    elliptic variational inequality
    0 references
    Banach space
    0 references
    Yoshida approximation
    0 references
    algorithm
    0 references
    convergence
    0 references
    numerical examples
    0 references
    performance
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references